Synthetic aperture radar
Radar sees at night. Now your COP does too.
Drop a Sentinel-1 SAFE, TerraSAR-X, ALOS or a Capella / ICEYE / Umbra SICD into the same ImageServer as your EO collects. Amplitude, speckle filtering, phase, real interferometric coherence and polarimetric decomposition run in pure C#; add ESA SNAP for terrain-corrected σ⁰, and SNAP + Snaphu (air-gapped) or ASF HyP3 (cloud) for a full InSAR displacement chain with PSInSAR / SBAS time series through MintPy.
Sentinel-1 GRD / SLCTerraSAR-X · PAZALOS PALSAR · ERS · Radarsat-1Capella · ICEYE · Umbra SICD6 speckle filtersInSAR: SNAP + Snaphu or HyP3MintPy time series
At a glance
- SensorsSentinel-1 (SAFE, SIGMA0 / BETA0 / GAMMA calib), TerraSAR-X / PAZ (TSX), SAR_CEOS (ALOS, JERS-1, ERS, Radarsat-1), NITF SICD / SIDD
- Pure-C# enginesAmplitude dB, Lee / Enhanced Lee / Refined Lee / Frost / Gamma / Lee-Sigma speckle, phase, coherence, Pauli RGB, Cloude-Pottier H/A/α, Freeman-Durden, Yamaguchi
- SNAP sidecarApply-Orbit → Thermal-Noise-Removal → Calibration → Speckle → Range-Doppler Terrain-Correction → dB; TOPSAR split + deburst for SLC
- InSARWSL backend: Back-Geocoding → ESD → Interferogram → TopoPhaseRemoval → Goldstein → Snaphu unwrap → TC → displacement COG; HyP3 backend with Earthdata auth
- Time seriesPSInSAR / SBAS via MintPy sidecar and HyP3 time-series scenes
- SICDTwo-tier reader — GDAL fast path + pure-C# NITF carrier + SICD XML fallback; phase-preserving complex I+Q
Capabilities
From backscatter to displacement
Amplitude and speckle
SAR rasters render as calibrated dB backscatter through the standard exportImage route — 10·log10(|z|² + ε), −25…−5 dB stretch — with six speckle filters selectable as rendering templates.
- sar-amplitude-db · sar-speckle-lee / -gamma / -refined-lee templates
- Enhanced Lee, Frost, Lee-Sigma in the filter engine
- GRD renders immediately; SNAP adds rigorous terrain correction
Phase and coherence
Interferometric phase on a cyclic HSV ramp and real coherence γ = |Σ(master × slave*)| / √(…) from a master / slave pair — the change-detection primitive for SAR.
- sar-phase · sar-coherence-real templates
- Coherence + amplitude change over an AOI
- Pair selection by SlaveCogId on the master collect
Polarimetric decomposition
Pauli RGB (R = |HH−VV|, G = 2|HV|, B = |HH+VV|) plus Cloude-Pottier H/A/α eight-zone, Freeman-Durden and Yamaguchi decompositions, computed in pure C# on full quad-pol data.
- PolarimetryEngine.cs — no external toolbox
- Dual-pol Sentinel-1 renders Pauli with the volume channel floored, and says so
InSAR displacement chain
Two backends behind one route: an air-gapped SNAP + Snaphu pipeline running under WSL2 or Docker, and ASF HyP3 in the cloud with a real Earthdata token flow. Output is an unwrapped displacement COG registered to the map.
POST /arcgis/rest/services/sar/insar/interferogram {backend:"wsl"|"hyp3"}- Copernicus 30 m, SRTM 3″ or GETASSE30 DEMs
- PSInSAR / SBAS time series via the MintPy host
SICD from commercial SAR
Capella, ICEYE and Umbra SICD products open through a two-tier reader: the GDAL NITF fast path when available, otherwise a pure-C# parser of the NITF carrier and SICD XML — phase-preserving, complex I+Q bands intact.
- Feeds the same amplitude, phase and coherence engines
- SIDD detected images render directly
SAR in the maritime picture
Vessel detections from SAR and EO (YOLOv8-OBB / RT-DETR sidecar, CFAR fallback) associate with AIS via Mahalanobis gating, Hungarian assignment and JPDA to surface dark-vessel candidates with evidence attached.
Analyst workflow
A change-detection pass in five steps
The pure-C# path needs nothing installed; the SNAP and Snaphu steps are optional sidecars that upgrade geometry and add unwrapped displacement.
- IngestRegister the master and slave scenes — a SAFE zip, TSX product or SICD — exactly like a COG.
- CalibrateRender calibrated dB immediately, or send the pair through the SNAP gpt chain for σ⁰ with Range-Doppler terrain correction (5–20 minutes per IW GRD).
- FilterApply Lee, Refined Lee, Frost or Gamma speckle filtering as a rendering template; compare in the swipe widget.
- CompareCompute real coherence between the pair; low-coherence and amplitude-change pixels flag activity, construction or damage.
- UnwrapRun the InSAR chain (WSL / Docker or HyP3) to an unwrapped displacement COG, then stack scenes for PSInSAR / SBAS through MintPy.
Specifications
Sensors, engines, sidecars
Inputs
- Bundled GDAL driversSAFE (Sentinel-1 GRD / SLC), TSX (TerraSAR-X / PAZ), SAR_CEOS (ALOS PALSAR, JERS-1, ERS, Radarsat-1), NITF (SICD / SIDD)
- Accept list.safe · .zip · .dim · .par · SICD NITF
- CommercialCapella · ICEYE · Umbra SICD two-tier reader
Engines (pure C#)
- AmplitudedB with ε guard, −25…−5 dB default stretch
- SpeckleLee · Enhanced Lee · Refined Lee · Frost · Gamma · Lee-Sigma
- Phase / coherenceCyclic HSV phase; real γ from master / slave
- PolarimetryPauli · Cloude-Pottier H/A/α (8-zone) · Freeman-Durden · Yamaguchi
Sidecars
- ESA SNAP gptRTC + calibration; separate ESA install (~0.6–1 GB); 30-min job timeout
- SnaphuUnwrapping under WSL2 / Docker (no native Windows binary)
- ASF HyP3Cloud InSAR; NASA Earthdata login
- MintPyPSInSAR / SBAS time series
- Route
POST …/ImageServer/sar/process
Honest scope
What ships, what is optional, what is next
Defense buyers get burned by overclaims. This is the plain-language scope for this capability, drawn from the product’s own documentation.
- Shipped Amplitude, speckle, phase, real coherence, polarimetry, SICD — Pure C# in the engine core; SAR Phases 1–3 and the polarimetric decompositions are in the build.
- Shipped InSAR chain with two backends — SNAP + Snaphu (air-gapped) and HyP3 (cloud) both ship; MintPy time series ships as a sidecar host.
- Optional sidecar / gated SNAP, Snaphu, HyP3, MintPy — All optional and capability-gated: SNAP is a separate ESA install; Snaphu runs under WSL2 or Docker; HyP3 needs an Earthdata account. GRD-only geometry is approximate until SNAP terrain-corrects it.
- Optional sidecar / gated Quad-pol required for full decomposition — Sentinel-1 is dual-pol, so the volume channel floors out; use quad-pol sensors for H/A/α, Freeman-Durden and Yamaguchi.
- Roadmap ISCE2 topsStack — The ISCE2 stack processor remains roadmap; MintPy consumes HyP3 and SNAP stacks today.
- Roadmap Commercial SAR tasking — Capella / ICEYE / Umbra tasking clients are built but not validated against live accounts — say “tasking-ready”; demo with archive data.
GeoTactical Insights · powered by 3D Map Explorer
See it on a disconnected machine.
A 30-minute guided demo on your own data: ingest, exploit, fuse, share to TAK and export — with the network cable out. We will also tell you exactly what is a sidecar, what is gated and what is still roadmap.
GeoTactical Insights
GEOINT and defense workflows powered by 3D Map Explorer, the air-gapped geospatial intelligence platform from Tech Maven Geospatial.